Did Computer Failure Bring Down Air France 447
<P>Did Computer Failure Bring Down Air France 447</P><P>**** Hidden Message *****</P> Diatribe 188 July 09<BR>1<BR>Diatribe 188<BR>Planely disastrous. Did Computer Failure Bring Down Air France 447?<BR>Speculation abounds following a plane crash. Mechanical flaws, terrorism, pilot error, and<BR>weather are the usual suspects. But in the tech age, where even your toaster is digital, IT<BR>systems must be added to the list. In the Air France disaster, there's a particularly urgent<BR>need for government authorities to eye the aircraft's on-board computer system as a possible<BR>culprit.<BR>That’s a quote from a US blog dealing with the recent crash of a an Air France 447,<BR>an Airbus 330 on June 3, off the coast of Mexico. 228 souls, men, women and children,<BR>passengers and crew, perished, and no-one survived. At the time, the aircraft was largely<BR>under the control of ADIRU, an electronic inertial reference system which interacts with<BR>the auto-pilot<BR>By way of contrast, earlier, on Jan.15th, another Airbus, at that time a 320, ditched in<BR>New York’s Hudson river after birds hit both engines which died, and nobody was even<BR>injured let alone lost. Control of that plane was manual, having not yet reached an altitude<BR>where the computers take over, in the hands of Capt. “Sully” Sullenberger, a former U.S.<BR>Air Force fighter pilot, former safety chairman of the Air Line Pilots Association, scholarly<BR>author on aviation safety, graduate of the U.S. Air Force Academy, and recently appointed<BR>Visiting Scholar at the University of California, Berkeley. Sullenberger was ably<BR>assisted by a crew who not only carefully instructed the passengers on what to do, but<BR>safely guided them to their rescuers who had been summoned by radio. Sullenberger,<BR>after executing a perfect “landing” on the river, topped his performance by twice walking<BR>the length of the half-submerged aircraft to make sure that all passengers were out.<BR>The majority of the blogs relating to the Mexican disaster were clearly lodged by industry<BR>representatives, insisted that at this time it was not yet possible to draw any conclusion<BR>regarding the latest event. From this they also decided that currently nothing can<BR>be done about it. If I were an intending passenger on an Airbus, I would not be likely to<BR>agree with their suggestion, so much the more so as there were accidents with our very<BR>own Airbuses too.<BR>One which reached prominence happened to a Qantas Airbus 330. Last year it<BR>started porpoising wildly while at cruising altitude, from memory somewhere over Western<BR>Australia. 51 passengers were injured, with damage ranging from broken bones to<BR>spinal trauma.<BR>The accident report issued by the Australian Transport Safety Bureau tells us: About<BR>two minutes after the initial fault, the air data inertial reference unit (ADIRU) generated<BR>very high, random and incorrect values for the aircraft’s angle of attack.<BR>Apart from this near crash, there were other warnings. Also last year, the US Federal<BR>Aviation Administration issued an airworthiness directive warning airlines about an “unsafe<BR>condition” associated with ADIRUs aboard Airbus 319, 320 and 321 models. The<BR>directive warned that the equipment was sending out bogus navigational fault warnings<BR>that could “result in loss of one source of critical altitude and airspeed data and reduce<BR>the ability of the air crew to control the airplane”. Earlier direct transmissions to Paris<BR>from the craft had indicated danger signals from the ADIRU system. Indeed, later evidence<BR>showed that the crew had actually attempted to turn the plane back before the<BR>crash.<BR>Diatribe 188 July 09<BR>2<BR>In 1990, while the Australian Hawke “Labor” government conspired with the tycoon<BR>Peter Abeles to lock out the experienced pilots of this country who were, amongst other<BR>things, demanding a say in air safety, I happened to be travelling to Sydney on unavoidable<BR>business sitting next to one of the locked-out captains who told me then about the<BR>difficulty of interfacing the human crew with what was then just the auto pilot. As a systems<BR>designer of sorts myself, this was not news to me.<BR>The technical problems are these: Pilots’ decisions regarding aircraft control are,<BR>amongst other things, based on instrument readings, control responses and control tower<BR>instructions. In the normal course of flying a plane, these details are acquired as you go<BR>along. If, on the other hand, you have to hand over control from machine to human, all<BR>this has to be done over a short period if not instantly. If this happens at a time of stress,<BR>mistakes are even more likely than if it happens in the normal course of events.<BR>Once upon a time, pilots flew planes by the use of hands and feet on column and<BR>pedals. With to-day’s 200 ton monsters, control surfaces are difficult to operate this way.<BR>Also, steel cables and hydraulic lines are heavy, and have to be duplicated or triplicated<BR>for safety. All that lifting capacity can be diverted to lifting passengers and cargo by<BR>operating controls and surfaces independently. The Airbus series of planes is therefore<BR>flown, as they say, by wire, using electrics or hydraulics. The human pilot’s commands<BR>are fed into the computer and the computer passes these commands on to the plane’s<BR>control surfaces, if it wants to, after subjecting the computer’s decision to the scrutiny of<BR>sundry mathematical processes. As a sop to the human pilot some pressures are simulated<BR>on pedals and stick to make the slave at the controls feel that he or she is in charge.<BR>Boeings are similar, but there at least the option exists for the pilot to exercise the ultimate<BR>control should the computers fail. No such options exist for the Airbus. The electronics<BR>can over-ride the human pilot.<BR>Spare a thought or two for him or her. World-wide, pilots would know what happened<BR>in Western Australia where they had to wrestle with the Airbus controls while passengers<BR>had their bones broken, and where a major disaster was avoided by a whisker. So far,<BR>the only steps taken to avoid disasters like the Mexican is to duplicate, triplicate or quadruplicate<BR>the faulty system.<BR>Every pilot, no matter how experienced, currently flying an Airbus relying on ADIRU<BR>– as most of them appear to be – would have to remember that they are at the mercy of a<BR>machine which has proved faulty and undoubtedly will prove so again It may well result in<BR>a situation which is beyond any pilot’s control. And since Abeles’ victory, aircraft captains<BR>cannot even refuse to fly the faulty planes. All this would do wonders for morale.<BR>There used to be a saying at Telstra that nowadays telephone exchanges were<BR>staffed by a man and a dog. The man is there to feed the dog and the dog is trained to<BR>stop the man from getting to the equipment. This fly-by-wire seems to follow a similar<BR>philosophy. Because the flight computers are programmed with the parameters which<BR>represent the limits of what the plane can do, they are made the masters of the human<BR>pilot.<BR>I remember a paper by an Indian academic which argued that in order to create safe<BR>systems they had to have a maximum of sophisticated technology backed up by the best<BR>available staff training. This is a nonsense. Good staff training can only be achieved by<BR>allowing staff maximum interaction to permit familiarity with the system. This was amply<BR>proved in disasters like Three Mile Island nuclear power station where human operators<BR>were given a few seconds to deal with a mass of complex problems after the automatic<BR>controls gave up. So did the operators.<BR>Step aside, Sully Sullenberger, we can no longer afford the likes of you at the aircraft’s<BR>controls. 法航的空难？ 好东西谢谢楼主
页:
[1]